2 50 SYSTEMS OF PERMANENT AGRICULTURE 



comparison, will, as a rule, give more trustworthy results. Thus, 

 we may subtract the average yields of the adjoining untreated 

 plots, 14 and 17, from the yields of the manured plots, 15 and 16, 

 to determine the increase produced by manure alone. Then we 

 may subtract the yield either of plot 4 or the average yield of the 

 three untreated plots, i, 4, and 7, from the yields of the plots 

 which receive both manure and phosphorus, to determine the in- 

 crease produced by manure and p'hosphorus combined, subtract- 

 ing from these figures the increases for manure alone, to determine 

 the effect of phosphorus. Still another method would be to average 

 all of the untreated plots whose results are in satisfactory agree- 

 ment, and discard the results of those that differ so widely as to 

 be clearly abnormal. By this method probably the results from 

 plots Ai, 67, and Ci would be discarded. 



By any of these methods of comparison, direct or indirect, it 

 will be found that, as a general average of all tests on all series, the 

 raw phosphate has produced practically the same gross increase 

 as the acid phosphate, although the acid phosphate applied cost 

 twice as much as the raw phosphate. 



Yet another indirect method of comparison can be made and 

 this one is preferred by the Ohio Experiment Station. This method 

 assumes that naturally the land varies somewhat uniformly from 

 one untreated plot to the next untreated plot, so that plot 15, for 

 example, if it had remained unmanured, would have produced a 

 yield equal to the sum of two thirds of the yield of plot 14 plus 

 one third of the yield of plot 17, and that this computed yield for 

 plot 15 (untreated), subtracted from the actual yield of plot 15, 

 gives the increase produced by the manure. The effect of the ma- 

 nure and phosphate is computed in the same manner, and the 

 difference gives the effect of the phosphorus. 



This method would be correct if the assumption upon which it is 

 based were correct; but considering that the change in the direc- 

 tion of such a curve is just as likely to occur on any other plot as 

 on the plots that happen to be numbered i, 3, 4, 7, etc., its appli- 

 cation may be of questionable value. However, in Tables 37, 

 38, and 39, the actual yields are reported for the twelve years, 

 and from these any one can make his own deductions. 



The total value of the three crops, based upon the average yields, 



